Untar specific files inside a .tar file

Hello,

I have a .tar file with over 100 files inside, and I only want to untar 10 of these files (I know their names). The untar function implemented in MATLAB untars all the files contained in the .tar file, and this would be a waste of time with so many files inside as well as if I need to untar other .tar files after it.

My question is similar to the one asked 8 years ago in the post 'untar selcted files from archive'. Is there any way to extract only some especific files inside the .tar file?

I have tried the aproach comented by Walter Roberson, but I think I am doing something wrong. If someone can explain me it a bit more in detail, or come up with another solution and share it, I would appreciate it.

Additionally, I would like to avoid messing up with the original untar function, if posible, and create a new one. I have tried to create a function called myuntar with the same code as the untar function, but it it fails as it is not able to reach the necessary functions the untar function is capable of, and the way to solve it is out of my knowledge.

Thank you in advance.

Answer 1

[Mohammad Sami]

You will have to use java for this.

A minimum working example is as follows (tested on R2020b).

tarpath = fullfile(pwd,'abc.tar'); % change here
file = java.io.File(tarpath); 
fis = java.io.FileInputStream(file);
tis = org.apache.commons.compress.archivers.tar.TarArchiveInputStream(fis);
copier = com.mathworks.mlwidgets.io.InterruptibleStreamCopier.getInterruptibleStreamCopier;
extractfolder = fullfile(pwd,'extracthere'); % change here
while true
entry = tis.getNextEntry;
if(isempty(entry))
break;
end
name = char(entry.getName)
if(endsWith(name,'.m')) % change the logic on which files to extract here
path = regexprep(name,'^[\.\\\/]*',''); % remove leading ..\ or ../
path = fullfile(extractfolder,path);
[folder,~,~] = fileparts(path);
mkdir(folder);
outJavaFile = java.io.File(path);
outStream = java.io.FileOutputStream(outJavaFile);
copier.copyStream(tis, outStream);
outStream.close();
end
end
tis.close;

[Ernesto Barbazán]

Thank you very much, this works like a charm. I just have one question about line 14, the one that says:

path = regexprep(name,'^[\.\\\/]*',''); % remove leading ..\ or ../

I do not know what do you mean with that comment, as the code seems to work for me without deleting any bar. I have tried to deduce it by looking at the documentation for the regexprerp function but I can not figure out what the second term as it is expressed ('^[\.\\\/]*' )exactly means.

Nevertheless, thank you again.

[Walter Roberson]

In ^[\.\\\/]* the ^ part means "match at beginning of string"; if the "lineanchors" option had been given, it would mean "match at the beginning of any line".

The * part means "any number (including 0) repetitions of the proceeding pattern.

The [] operator means to match any one character that is listed inside the [].

The \. means a literal period. The \\ means a literal backslash. The \/ means a literal forward slash.

So the [] construct is standing for "either a period or a backslash or a forward slash".

It happens that inside [] that period and forward slashes do not have special meaning, so the construct could have been written as [.\\/] instead.

Start at the beginning of the line. Look for any number of period or slashes there. Replace them with '' (emptieness).

[Ernesto Barbazán]

Ok, I think I get it now. Thank you for the detailed explanation.

[Mohammad Sami]

The line is to remove ../ or ..\ from the beginning of the path.

This guards against a slip attack, where someone can extract contents into system directory by appending multiple ..\ in front of the path. Whent the fullfile function resolves the path, it will end up resolving to a system folder.

https://github.com/snyk/zip-slip-vulnerability